WCB Cost Model Virtual Workshop 2012 -Operating Expenses Input Values

Please provide comments to the issue below as part of the 2012 WCB cost model virtual workshop for inclusion in the record. Comments are moderated for conformity to the workshop’s guidelines.

Background

The Connect America Cost Model (CACM) calculates monthly operating expenses (opex) generally as a dollar amount representing a percentage of capital investment or a percentage of assumed average revenue per user (ARPU). Here we describe in more detail the methodology used to develop opex values and seek comment on whether any adjustments should be made to the specific input values currently contained in CACM version 3.0.

Parties who have signed the Third Supplemental Protective Order may view the CACM version 3.0 opex input values, along with a more detailed description of how the inputs were calculated and supporting spreadsheets, by accessing the model, visiting the Resources page, and opening the Opex Overview file.

CACM uses publicly-available data where possible for opex input values, supplemented by other sources where there is no readily-available public source of information for opex, to develop baseline opex amounts per month per subscriber line, by company size. Opex input values vary depending on classification of the company size at the holding company level as described in the OCNCoSize input table.

Opex values also vary depending on the classification of the density of the study area as “urban,” “suburban,” and “rural.” The classifications are based on NECA’s classification of the study area code. Because NECA does not utilize a suburban classification, CACM uses the same inputs for suburban densities as those used for urban densities.

To calculate the Network Operations Expense input values:

• CACM primarily utilizes network operating cost data from National Exchange Carrier Association (NECA) data (pulled from “Universal Service Fund Data: NECA Study Results”) from 2008 to 2010; ratios of opex to capex were derived to create a large company baseline for each category of expense and the results were validated against company-specific data provided by ABC Coalition Company members.
• The network opex factors were then adjusted to reflect a forward-looking network topology by applying a ratio of modeled capex to booked capex (based on ARMIS and NECA data).
• Once model output was available, the scaling was revisited to ensure that forward-looking opex values did not exceed NECA-based booked opex.
• Using knowledge of industry costs, the cable factors in the Network Opex were further segregated between metallic and non-metallic to account for the operating cost differences between the two types of cable.
• Finally, the opex factors were applied to the capex estimates in the model to determine monthly opex input values for network operations expense.

To develop CACM General and Administrative (G&A) input values, a regression of five years (2006 - 2010) of NECA G&A Opex (dependent variable) and Total Plant in Service (TPIS) (independent variable) data segregated by company size was performed to determine the relationship between total plant investment and G&A operating expenses, with a forward-looking to historical G&A adjustment factor applied by company size and by density; again, the results were validated against company-specific data provided by ABC Coalition Company members.

CACM Customer Operations and Marketing input values were developed using Automated Reporting Management Information System (ARMIS) data for large and mid-sized incumbent local exchange carriers (pulled from “FCC Report 43-01, ARMIS Annual Summary Report”). ARMIS data are not available for all carriers and have been discontinued over time for others. (For example, starting with the 2008 reporting year, Verizon, AT&T, and Qwest were no longer required to file FCC Report 43-01.) 2007 and 2010 ARMIS data were used to determine the ratio of customer operations and marketing expense to revenue and validated against 2010 company-specific data provided by ABC Coalition Company members; CACM then applies those resulting percentages to an assumed ARPU to develop a per line dollar amount input value. The use of the most recently available historical data is consistent with the approach taken by the Commission in developing input values for the Hybrid Cost Proxy Model.

CACM calculates bad debt input values by assuming an industry-standard of 2 percent of ARPU.

The overwhelming majority of data supplied to NECA and ARMIS are for networks with a twisted-pair-copper last mile. For comparison purposes, we provide below the per-subscriber network operating input values from version 3.0 of CACM for FTTd and indicate how those input values compare to NECA and ARMIS data for opex. Note that ARMIS data include operating expenses for providing all services, including special access and private line services, which are excluded from cost reporting in the CACM; one therefore would expect the CACM estimates to be lower than the ARMIS data. We then provide a comparison between CACM version 3.0 FTTd and fiber-to-the-premises (FTTP) network operating costs per subscriber per month; one would expect the FTTP opex input values to be lower. Finally, we provide a per subscriber summary of the non-network operating costs for both FTTd and FTTP; the G&A values are lower for FTTP than FTTd.

Per-subscriber costs vary depending on assumptions regarding likely take rate: network opex, calculated as a weight on the capital investment, remains largely constant regardless of take rate, so that the per-subscriber cost will be higher if one assumes fewer subscribers. To illustrate how opex input values will vary depending on assumed take rate, we provide below various CACM version 3.0 input values at differing take rates – specifically, at a 90 percent and 80 percent take rate.

CACM v3.0 FTTd network opex per subscriber per month ($) compared to NECA network opex per line per month by company-size category

	Large	Medium	Small	X-Small	XX-Small
CACM FTTd network opex per sub (90%)	$10.69	$10.04	$28.39	$35.24	$45.18
CACM FTTd network opex per sub (80%)	$11.85	$11.15	$31.77	$39.48	$50.62
NECA network opex per sub	$13.24	$14.14	$27.79	$39.39	$51.63

 

CACM v3.0 FTTd network opex per subscriber per month ($) compared to ARMIS 43-01 data

	Large	Medium	Small	X-Small	XX-Small
CACM FTTd network opex per sub (90%)	$10.69	$10.04	$28.39	$35.24	$45.18
CACM FTTd network opex per sub (80%)	$11.85	$11.15	$31.77	$39.48	$50.62
ARMIS network opex per sub	$21.36	$21.36	NA	NA	NA

 

Comparison between CACM v3.0 FTTP and FTTd network opex per subscriber per month

	Large	Medium	Small	X-Small	XX-Small
CACM FTTP network opex per sub (90%)	$6.16	$6.72	$19.61	$23.84	$31.47
CACM FTTd network opex per sub (90%)	$10.69	$10.04	$28.39	$35.24	$45.18
Ratio (FTTP/FTTD)	58%	67%	69%	68%	70%
CACM FTTP network opex per sub (80%)	$6.62	$7.30	$21.75	$26.54	$35.01
CACM FTTd network opex per sub (80%)	$11.85	$11.15	$31.77	$39.48	$50.62
Ratio (FTTP/FTTD)	56%	65%	68%	67%	69%

 

Per Subscriber summary of G&A, Customer Operations, Sales, and Marketing, and Bad Debt operational costs in CACM V3.0

		Large	Medium	Small	x-Small	xx-Small
FTTd, 90% take	CustomerOperationsMktOpex	$5.93	$5.93	$5.51	$7.35	$9.95
	GenAdminOpex	$4.00	$6.02	$13.44	$28.17	$45.82
	BadDebtOpex	$0.92	$0.92	$0.92	$0.92	$0.92
	Total	$10.85	$12.87	$19.88	$36.44	$56.69
FTTd, 80% take	CustomerOperationsMktOpex	$5.93	$5.93	$5.51	$7.35	$9.95
	GenAdminOpex	$4.41	$6.68	$15.04	$31.56	$51.36
	BadDebtOpex	$0.92	$0.92	$0.92	$0.92	$0.92
	Total	$11.26	$13.53	$21.47	$39.84	$62.23
FTTp, 90% take	CustomerOperationsMktOpex	$5.93	$5.93	$5.50	$7.35	$9.95
	GenAdminOpex	$3.93	$5.91	$13.02	$27.49	$44.82
	BadDebtOpex	$0.92	$0.92	$0.92	$0.92	$0.92
	Total	$10.78	$12.76	$19.44	$35.77	$55.69
FTTp, 80% take	CustomerOperationsMktOpex	$5.93	$5.93	$5.50	$7.35	$9.95
	GenAdminOpex	$4.25	$6.47	$14.48	$30.67	$50.05
	BadDebtOpex	$0.92	$0.92	$0.92	$0.92	$0.92
	Total	$11.10	$13.32	$20.90	$38.95	$60.92

Questions for Comment

1. Is it appropriate to assume that 80 percent of locations will subscribe over the five-year time horizon as we determine the appropriate input values for opex? (This assumes 60-70 percent of locations subscribing at any one point in time, with some churn over time.) If the Bureau ultimately adopts a take rate different from 80 percent, should it scale the opex inputs to maintain a constant opex per subscriber, or allow opex per subscriber to change as seen in the tables above?
2. Although few carriers are deploying FTTP networks widely, available data indicate that network operating expenses for FTTP networks are substantially lower than those for copper-based networks. As shown in the charts presented above, the network opex input values in CACM version 3.0 for FTTP are 1/3 to 1/2 less than the input values for FTTd. Is that cost differential appropriate?
3. CACM version 3.0 currently uses a $46 ARPU (for broadband only) in developing the bad debt and customer operations and marketing input values. The National Broadband Plan model estimated the ARPU of fixed voice service at approximately $33.50 and the ARPU of fixed broadband at $36-44. The total of the two services together would be roughly $70-80. We note that customers that subscribe to both services may get a discount, while others may subscribe to only one service. Given this, is it reasonable to assume a $65 ARPU when developing bad debt and customer operations and marketing input values? Are the values for bad debt of $1.30 per subscriber per month, and $8.43 for customer operations and marketing appropriate, instead of the input values shown in the table above?
4. CACM generally develops input values based on data for the large companies, and then applies factors to scale those inputs for companies of a smaller size. In an ex parte, ACA has argued that G&A costs are likely overstated because they are linearly related to capital expenses. It contends that the cost model fails to reflect that price cap carriers generally have operating leverage for G&A costs, and that incremental G&A expenses should go down as revenues increase. The charts above illustrate that the G&A costs per subscriber are much lower for the larger companies than the smaller companies. Are the input values for the large companies appropriately reflective of the relative efficiencies of larger scale operations?
5. Should the Bureau make any changes to specific opex inputs? To the extent commenters argue opex inputs should be adjusted, they should describe with specificity why the values should be adjusted and by what amount.